The military-industrial scientific research system of the academic master.

During the month when Chang Haonan was busy with the QC300 project, Torch Group was also dealing with the aftermath of the Conference on Numerical Algorithms and Computational Science.

First of all, of course, three different versions of TORCH Multiphysics 2.0 software were officially released:

In addition to the existing professional version and professional upgraded version, at the strong request of many users, Torch Group also launched the "pre-sale" of the Dawn version of the software in advance.

That is the version that is bundled with Sugon 3000T.

In fact, when Chang Haonan announced the Dawn version at the opening ceremony of the conference, he didn't think that many people would actually buy it.

After all, although the Sugon 3000T is compact in size among supercomputers, it is still a behemoth with 10 sets of cabinets and a weight of 5 tons.

For most small and medium-sized users, as well as foreign users, it is actually not convenient to purchase such a set of things.

In addition, as China's first supercomputer that truly reaches the international mainstream level, it has very stringent performance requirements for components. At the same time, as the leader of the current domestic semiconductor industry, Sugon Group needs to increase the localization rate of its products as much as possible.

High performance means high difficulty in production.

For the domestic semiconductor industry, which is just starting out, even if it uses the same imported equipment and raw materials as others, the yield rate is still not optimistic.

As a result, after the localization rate increased, the cost not only did not decrease, but increased instead.

Of course, this issue is secondary.

After all, Torch Group’s products have never been focused on cost-effectiveness, and its core user groups don’t care much about price.

The key is the lead time.

In the Dawning 1000 era, Beijing University of Aeronautics and Astronautics inadvertently set off a wave of colleges and universities purchasing supercomputers and servers.

However, at that time, the mainstream was still mainly imported foreign products, and Shuguang Group only got some benefits.

Even so, it still helped Shuguang Group, which had suffered huge losses for years, to turn losses into profits.

Avoiding being sold to Dream Group is secondary. What is more important is that for the first time, a complete loop of product research and development - market launch - generating revenue - continuing to invest in product research and development has been opened up.

By the dawn of the 2000s, due to the influence of the Qinghai-Tibet Railway and the West-East Gas Pipeline Project, this trend was inherited by many central enterprises in monopoly industries.

During this process, Sugon Group also accumulated a considerable number of users.

The result is that before the Sugon 3000 series supercomputer project passed the acceptance, more than 50 users, including Torch Group and Aviation Power Group, had placed orders for more than units.

It has not been delivered yet.

Therefore, the earliest delivery time for Sugon 2000T ordered now is the end of the year.

For many users, this cycle seems too long.

Taking all these factors into consideration, Chang Haonan initially planned to release only two versions, Pro and Pro plus.

As for the Dawn version, we will wait until it can be delivered at the end of the year.

However, the situation was far beyond his expectations.

In fact, with the increasingly close exchanges between China and Western countries in recent years, even the Pro plus version, which was previously only available to some domestic units, has gradually been opened up to foreign partners with sufficient specifications, such as Kompas, Tupolev, Snecma and Dassault.

But even so, there are still many users who cannot obtain Pro plus permissions or are discouraged by its outrageous price and have shown great interest in the Dawn version.

Although the Dawn version is sold as a bundle, the Dawn 3000T itself is a supercomputer with performance ranking among the top in the world, and the software is equivalent to a free gift, so it is not a loss at all.

It is even more cost-effective than buying just one Sugon 3000.

Before the two-day Algorithm and Computational Science Conference ended, the number of users who came to inquire about this version and clearly expressed their intention to order it was close to three digits...

Finally, unable to bear the excessive enthusiasm, Chang Haonan had to open the reservation channel in advance and promised to deliver the products in the order of reservation starting from the end of the year.

Compared with the colleagues who sold cars at press conferences before their rebirth, the only difference is the policy of "50,000 deposit offset 100,000 advance payment".

However, Chang Haonan is facing the situation that there are too many orders and he is worried that he will not be able to deliver all of them, so he has no motivation in this regard.

In short, after the conference, when Chang Haonan saw Academician Li of the Shuguang Group one day in early April, he had no doubt that if the age difference between the two was not so great, the latter would be able to perform a "Jie has been wandering for half his life, and only regrets that he has not met a wise master..." on the spot.

In addition to the release of the new version of the software, another thing is the Numerical Computing-Graphical Modeling Alliance jointly established by Torch Group, Dassault Systemes, and Kompas.

In the past, TORCH Multiphysics also supported direct import of models from various mainstream formats of modeling software.

However, anyone who has experienced the wild period of the software industry in the early 21st century knows it.

Compatibility does not mean it is easy to use.

Especially after Torch Group upgraded their mesh generation technology many times, some old 3D models began to show serious problems with local conditions.

As a result, it is necessary to specifically adjust the model building nodes before automatically generating the mesh.

but.

After the formation of this software alliance, the three companies will fully open their interfaces to each other to achieve maximum data compatibility.

In fact, this has been achieved in the new version of TORCH Multiphysics.

For users, the performance of the software itself has not changed.

But it has brought exponential improvement to the user experience and work efficiency.

The first project to benefit from this software alliance is exactly what Chang Haonan has been thinking about.

KJ-2000...

……

Almost at the same time as the QC300 project review meeting was held.

Jinling.

In a meeting room of the 14th Institute of Electronic Science and Technology.

A technical seminar on the design of KJ-2000 is being held.

It was called a technical seminar, but not many people attended.

There were less than 10 people sitting on both sides of the long conference table.

On one side are Academician Wang Xiaomo, chief scientist of the National Early Warning and Detection Project and general manager of the KJ-2000 project, and two other leaders from the General Armament Department and the Air Force Armament Department.

Opposite the three of them were Guo Lin, who had been promoted to deputy director of the 14th Institute and director of the antenna and microwave department, and chief designer of the radar system, as well as the heads of five other key subsystems, and an empty seat.

This seat originally belonged to Liang Shaoxiu, who was seconded to Gaofei Group as chief platform engineer.

He stood behind the desk from the beginning of the meeting, and then never found a moment to sit down and rest...

The reason for all this can be traced back to Liang Shaoxiu's first words after the opening:

"Dear leaders, our platform design team has successfully completed the three-dimensional reconstruction of the C909 aircraft itself and the disc-shaped radome, and has calculated the aircraft and the plan after the aircraft is equipped with the radome..."

Originally, this was just a routine monthly technical meeting.

But once this sentence was uttered, the nature of the matter was completely different.

"Comrade Shao Xiu, I remember... at the meeting last month, you said that this work would take about 4-6 months to complete?"

Wang Xiaomo and Guo Lin had witnessed this kind of situation where progress suddenly increased rapidly in the past.

But the one standing on the stage today was Liang Shaoxiu instead of Chang Haonan...

So it was a huge surprise indeed.

"That was exactly what happened last month."

The reason why Liang Shaoxiu chose to reveal this news during the meeting was because he was influenced by Chang Haonan. At this time, he finally experienced the pleasure of "shocking his peers for a whole year":

“Because the modeling and calculation work of the radome is not large, the focus is still on the aircraft.”

"In fact, we had received the aircraft design data from the Shanghai Aircraft Group before, but the data came from Tupolev. Although the annotations had been translated into Chinese, the file format was not fully compatible with the CATIA we were used to, so we estimated that it would take several months to correct the format."

“Importing the CATIA model into a numerical calculation program requires about the same amount of proofreading and modification work…Compared with these two processes, CFD and CSD calculations take up less time.”

"But after we actually started working on it, we found that solving the compatibility issue was not as difficult as we expected, so we released the results ahead of time..."

This explanation made the people sitting below at a loss for words for a moment.

After a while, Wang Xiaomo waved his hand, signaling Liang Shaoxiu to continue.

The latter did not waste any more words and continued to refer to Chang Haonan's habits.

Here is a heavy conclusion:

"From the results, it can be seen that even after the C909 aircraft platform is raised and equipped with two tail fins, it is still difficult to carry the ellipsoidal radome with a diameter of 3 meters and a thickness of 10.5 meters in the planned plan 1.9."

"When this radome is set at a distance of 2.8 meters from the fuselage, the downwash generated by the wing will increase, causing the angle of attack of the horizontal tail relative to the actual airflow to decrease, and the lift generated by the horizontal tail to decrease. The maximum effective lift coefficient of the entire aircraft will be reduced to 1.107..."

When he said this, the air in the entire conference room almost froze.

电科14所给空警2000预设了三个雷达方案，直径分别为10.5米、9米和8米。

Although the array area that can be accommodated by the same radome has increased by about 30% after the three-sided array was changed to a two-sided array, no one would think the antenna size is too large, especially when the packaging technology is relatively backward.

Therefore, the project team still regards the 10.5-meter radar dome as the priority option.

If the calculation results prove that it is not feasible, then there is a high probability that we will have to make a second-best choice.

However, Liang Shaoxiu did not let everyone worry for too long.

After a short pause, he changed the subject:

"However, we tried to adjust the shape of the radome itself without changing its cross-sectional area so that it could generate some lift, and then conducted new tests..."

As he spoke, he clicked the mouse, and the cross-section of the radar dome on the screen immediately turned into an ellipsoid that was slightly flattened at the bottom:

"After lift compensation, the influence of the radome on the maximum lift coefficient of the entire aircraft is reduced, from 1.193 in the clean configuration to 1.178, which fully meets the design requirements!"